Opportunity Description
Summary: Gemini 1.5 introduces a significant advancement in AI with its ability to process up to 2 million tokens, far surpassing the typical 32,000–128,000 token context window of earlier models. This breakthrough enables the handling of vast amounts of information, such as 100,000 lines of code or 10 years of text messages, making traditional methods like vector databases and RAG less crucial. Instead, in-context retrieval and many-shot prompting, where models are shown hundreds or thousands of examples, become more viable. Initial tests by Google DeepMind showcased Gemini 1.5’s impressive performance in long-document and long-video QA, and it even successfully created documentation and answered questions about a 1924 film, Sherlock JR. This competition encourages building creative applications of the model through Kaggle Notebooks and YouTube Videos.
Mode: Online
Eligibility: Open to individuals with knowledge of machine learning, data science, and programming
Fees: Free
Application Deadline: Dec 2, 2024, 23:59:59
Notes: Participants should have prior experience in machine learning, as the competition will involve advanced data analysis and model-building.
